Security audit
Security checks for vulnerabilities and agentic risk
This is a markdown-only sales lead qualification skill with clear scope and no executable code, network use, persistence, or credential access.
Install this if you want help evaluating active sales leads. Avoid pasting confidential customer details unless you are comfortable sharing them with the agent session, but this skill itself does not add external connections or hidden persistence.
Code or instructions that leak agent conversation context to external services, potentially exposing sensitive user interactions.
When user provides lead information, follow this sequence.
### Step 1: Extract context
Parse input for:
- prospect identity
- company and role
The trigger phrase set includes generic language such as "lead assessment" and similar common sales terms that could appear in ordinary conversation, making unintended activation plausible. In an agent skill system, overbroad invocation can cause the skill to engage when the user did not explicitly request it, which may steer workflow decisions or expose internal sales reasoning in the wrong context.
The manifest provides positive trigger phrases but no limiting conditions, exclusions, or negative examples, so the dispatcher has little guidance for distinguishing deliberate use from incidental mention. This increases the chance of accidental activation, especially in business chats where lead-qualification terminology is routine.
No suspicious patterns detected.